Imaging the formation of a p-n junction in a suspended carbon nanotube with scanning photocurrent microscopy

Imaging the formation of a p-n junction in a suspended carbon nanotube

with scanning photocurrent microscopy

Gilles Buchs,1,a)Maria Barkelid,1Salvatore Bagiante,2Gary A. Steele,1and Val Zwiller1 1

Kavli Institute of Nanoscience, TU-Delft, Post Office Box 5046, 2600 GA Delft, The Netherlands

2

Istituto per la Microelettronica e Microsistemi, Consiglio Nazionale delle Ricerche, Stradale Primosole 50, I-95121 Catania, Italy

(Received 13 July 2011; accepted 20 August 2011; published online 5 October 2011)

We use scanning photocurrent microscopy (SPCM) to investigate individual suspended semiconducting carbon nanotube devices where the potential profile is engineered by means of local gates.In situ tunable p-n junctions can be generated at any position along the nanotube axis. Combining SPCM with transport measurements allows a detailed microscopic study of the evolu-tion of the band profiles as a funcevolu-tion of the gates voltage. Here we study the emergence of a p-n and a n-p junctions out of a n-type transistor channel using two local gates. In both cases theI - V curves recorded for gate configurations corresponding to the formation of the p-n or n-p junction in the SPCM measurements reveal a clear transition from resistive to rectification regimes. The rectifi-cation curves can be fitted well to the Shockley diode model with a series resistor and reveal a clear ideal diode behavior.VC 2011 American Institute of Physics. [doi:_{10.1063/1.3645022]}

I. INTRODUCTION

The unique electronic properties of carbon nanotubes make them ideal systems for future large-scale integrated nanoelectronics circuits.1Due to their quasi-one-dimensional geometry, the electronic bands of carbon nanotubes can be engineered by means of electrostatic doping. In this context, p-n junction diodes2–7 as well as tunable double quantum dots working in the single particle regime have been realized in suspended nanotube devices using local gates.8High spa-tial control and resolution of the electrostatic doping of semi-conducting nanotubes will allow the realization of electronic and optoelectronic devices like diodes or phototransistors9 with tunable properties, which is not possible for devices based on chemically doped semiconductors. Moreover, a controlled confinement of single carriers in combination with a p-n junction10 in a semiconducting nanotube could potentially enable future applications such as electrically driven single photon sources in the burgeoning field of car-bon nanotube quantum optics.11

Here we report on a scanning photocurrent microscopy (SPCM) study of suspended semiconducting nanotube devi-ces where the band profile is engineered by means of local gates in order to generate p-n junctions at controlled loca-tions along the nanotube axis.

II. EXPERIMENT

The devices consist of a nanotube grown between plati-num electrodes over predefined trenches with a depth of 1 lm and widths of 3 or 4 lm. Up to four gates are defined at the bottom of the trenches. A schematic and a scanning elec-tron microscopy image of a typical device with four gates and a 3 lm wide trench are shown in Figs. 1(a) and 1(b), respectively. The fabrication began with a pþþ silicon wafer

used as a backgate covered by 285 nm of thermal silicon ox-ide. On top of this, gate electrodes made of 5/25 nm W/Pt were defined using electron-beam lithography, followed by the deposition of a 1100 nm thick SiO2 layer. A 1000 nm deep trench was dry etched, leaving a thin oxide layer on top of the gates. A 5/25 nm W/Pt layer was then deposited to serve as source and drain contacts, and nanotubes were grown at the last fabrication step at a temperature of 900C from patterned Mo/Fe catalysts.8,12

In SPCM, photocurrent (PC) is recorded as a laser spot is scanned across a sample. PC appears when photogenerated electrons and holes are separated by local electric fields in the device, such as those present at metal/nanotube interfa-ces,13,14 defect sites,14 or p-n junctions.5 Our SPCM setup consists of a confocal microscope with aNA¼ 0.8 objective illuminated by a k¼ 532 nm laser beam. The diffraction lim-ited spot is scanned using a combination of two galvo-mirrors and a telecentric lens system while the dc PC signal and the reflected light intensity are recorded simultaneously in order to determine the absolute position of the detected PC features. Typical light intensities of 3 kW/cm2are used in this work.

III. RESULTS AND DISCUSSION

Figure 1(c) shows the superimposition of the PC and reflection images of a device with two gates labeled G1 and G2 separated by 250 nm in a 4 lm wide trench, measured in vacuum ( 104 mbar) at room temperature. The applied voltages are VG1¼ VG2¼ þ8 V and VSD¼ 0 V. The PC image in combination with the measured transfer characteris-tics reveal the presence of a single semiconducting nanotube crossing the trench, whose axis is indicated by a dashed line. Figure 1(d) shows the PC line profile recorded along the dashed line in panel (c) with two PC spots at the edge of the trench revealing the local electric field generated at the metal/nanotube interface. The corresponding band diagram

a)_{Electronic mail: g.buchs@tudelft.nl.}

0021-8979/2011/110(7)/074308/4/$30.00 110, 074308-1 VC2011 American Institute of Physics

JOURNAL OF APPLIED PHYSICS 110, 074308 (2011)

is depicted in (e) with an illustration of the photogenerated carrier separation at the contacts. The asymmetry in the PC is due to different resistances for carriers at the source and drain contacts, for instance here a thinner Schottky barrier for electrons at the drain contact (D).13

In Fig.2, we demonstrate the imaging of a p-n junction, whose position and polarity can be tuned with the local gates. We use a four trench gate device similar to the one shown in Fig. 1(b). The measurements are performed at room temperature in air. With the same potential applied to all four gates, two PC spots appear at the trench edges with polarities depending on the gates potential, similar to those shown for the two gate device in Fig. 1(c). For opposite potentials (þ8 V/8 V) applied to groups of adjacent gates with configurations G1-G2, G3-G4 (panel (a)), G1-G2-G3, G4 (panel (b)), and G1, G2-G3-G4 (panel (c)), p-n junctions are created and clear PC spots appear at the electric field maxima corresponding to depletion regions. The images demonstrate that we can both move the position of the pn-junction and change its polarity using the local gates. Note that the metal/nanotube interface does not show PC signals due to the potential barrier formed at the depletion region that blocks one of the photogenerated carriers, as illustrated in panel (d). The patterns around the maximum intensity PC spots are due to diffraction effects from the structure of the gates. Gaussian fits to the PC signals along the nanotube axis (dashed lines) show that the center of the depletion regions is positioned close to the center of the spacing between two gates with opposite potentials.

In Fig. 3we study a two trench gates device illustrated in the schematic in Fig. 3(a) (4 lm wide trench and gate separation 250 nm). The measurements have been performed

at room temperature in vacuum. A single semiconducting nanotube was found to cross the trench establishing an elec-trical contact between source and drain electrodes. Using the technique described in Refs. 13 and 15, we estimate the bandgap to be Eg 400 meV, corresponding to a diameter of about 1.7 nm,16_{and we find that the Fermi level at the} con-tacts lies at about one third of the bandgap below the conduc-tion band.17A transition from a fully n-type or n-n channel to p-n (n-p) configuration is studied in panels (a)–(c) ((d)–(f)) by applying a constant potential ofþ 8 V to G2 (G1) and sweeping G1 (G2) fromþ 8 V to 8 V. For each value ofVG1(VG2), the laser spot is scanned along the nano-tube axis and the PC is recorded, Fig. 3(b) (Fig.3(e)). For the range VG1(VG2) 0 V, the PC shows two contributions at the Schottky barriers. Below 0 V the negative (positive) PC signal starts to move its position toward the center of the trench and the positive (negative) PC signal vanishes.

Both effects are due to a transition of the band profile from pure n-n to the configuration depicted with the label i-n FIG. 1. (Color online) (a) Schematic of a device with four trench gates

G1-G4. A diffraction-limited laser spot (k¼ 532 nm) is scanned across the device and PC is recorded between source (S) and drain (D) contacts. (b) Scanning electron microscope image of a four trench gates device. (c) Superimposition of the PC image and the reflection image for a device with two trench gates separated by 250 nm, measured with VG1¼ VG2¼ þ8 V

and VSD¼ 0 V. A single semiconducting nanotube is highlighted with a

dashed line. (d) PC line profile recorded along the dashed line in panel (c) corresponding to the nanotube axis. (e) Corresponding band diagram with photogenerated carrier separation at the metal/nanotube interfaces.

FIG. 2. (Color online) Tuning the position and polarity of a p-n junction using local gates. (a)–(c) Left column: Superimposition of the PC and reflec-tion images and the configurareflec-tions of the potentials applied to the trench gates. The dashed line corresponds to the nanotube axis. Right column: Cor-responding band diagrams with PC and reflection intensities (R) recorded along the dashed line as well as the position of the gates. Each PC line pro-file is fitted with a Gaussian discarding the diffraction-induced patterns. (d)–(e) Band diagrams illustrating the behavior of the photogenerated car-riers at different positions: metal/nanotube interfaces and depletion region, respectively.

074308-2 Buchs et al. J. Appl. Phys. 110, 074308 (2011)

(n-i) where the drain (D) (source (S)) side of the n channel begins to pinch off and prevents electrons generated at the source (S) (drain (D)) Schottky barrier from reaching the drain (source) contact. The negative (positive) PC signal continues to move into the trench until it is suppressed below VG1¼ 1 V(VG2¼ 0.8 V) and then recovers around VG1¼ 3 V (VG2¼ 2 V). This low PC intensity likely indi-cates a shallow potential profile in which the electric field is not large enough to separate the photogenerated carriers. At VG1 3 V (VG2 2.5 V), the PC signal increases drasti-cally up to about22 pA (30 pA) and shifts slowly toward the center of the trench atVG1(VG2)¼ 8 V. This strong PC signal is the consequence of hole doping of the drain (source) side of the device, resulting in a p-n (n-p) junction with a large

elec-tric field in its depletion region,13 depicted in the band diagram corresponding to VG1(VG2)¼ 8 V in panel (b) (panel (e)).

In addition to PC imaging, we also performI - V meas-urements (dark current) for values of VG1 (VG2) indicated by labels 1-7 (10-60) in panel (b) (panel (e)). A progression from ohmic regime at VG1 (VG2)¼ þ8 V with a measured resistance of about 40 MX (68 MX) (Ref. 18) to a clear rectification behavior starting below VG1(VG2)¼ 2 V cor-responding to I - V curves 4-7 (40-60) with the forward cur-rent increasing with jVG1j ( Vj G2j) is observed. The

rectification curves 4-7 ((40-60)) can be fitted well to the Shockley diode model I¼ I0ðeVSD=ðnVTÞ 1Þ with a series

resistor,3

FIG. 3. (Color online) Imaging the emergence of a p-n junction. (a) Schematic of a device with two trench gates. G2 is set toþ 8 V and G1 is swept. (b) PC transition map recorded along the nanotube axis, with corresponding band diagrams for n-n, i-n, and p-n regimes. (c) I - V curves recorded at values of VG1

la-beled 1-7 in (b). Inset: Values of the series resistances R1 for the linear n-n regime and R4- R7 for the rectification regime fitted with the Shockley diode model. (d) G1 is set toþ 8 V and G2 is swept. (e) PC transition map recorded along the nanotube axis, with corresponding band diagrams for n-n, n-i, and n-p regimes. (f) I - V curves recorded at values of VG2labeled 10-60in (e). Inset bottom right: Values of the series resistances R10for the linear n-n regime and

R40-R60for the rectification regime fitted with the Shockley diode model. Inset up left: Transfer characteristics of the device, where VG1-G2corresponds to the

voltage applied simultaneously to G1 and G2. a is the estimated gate efficiency.

074308-3 Buchs et al. J. Appl. Phys. 110, 074308 (2011)

I¼ I0 nVT I0R W I0R nVT e VSDþI0R nVT    1   ; (1)

where I0 is the saturation current at reverse bias, n is the ideality factor,VTis the room temperature thermal voltage of 26 mV,R is the series resistance, W is the Lambert W-func-tion, andVSDis the source-drain voltage. For a measured sat-uration current of aboutI0¼ 4  1013A (I0¼ 4  1013A), we find the best fit withn¼ 1 and R4 ¼ 25 GX, R5 ¼ 2 GX, R6¼ 690 MX, and R7 ¼ 200 MX (R40_{¼ 25 GX, R5}0_{¼ 4 GX,}

and R60¼ 1.2 GX). The decreasing value of R with Vj G1j

(jVG2j) is in good agreement with the band profiles model

depicted on the right side of panel (b) (panel (e)), implying a decrease in width of the tunneling barrier for hole injection in the segment of the nanotube located above G1 (G2) when

VG1

j j ( Vj G2j) increases. The higher current in forward bias

for the p-n configuration compared to n-p is due to an asym-metry in the resistance at the source and drain contacts. We note that the estimated bandgap from the turn-on voltage (VSD 150 mV) is not consistent with the value estimated from the transfer characteristic (Eg 400 mV). In addition to the systematic error from the estimation of the bandgap from the transfer characteristic using the method of Refs.13

and15, this difference can also potentially be due to diffu-sion effects for the carriers, recently observed by another group.7Such discrepancies in our devices will be the subject of future investigations.

IV. CONCLUSIONS

In summary, we have demonstrated the control of the position and polarity of a p-n junction in multigate sus-pended carbon nanotube devices. We created a p-n junction from a purely n-type channel and imaged its formation using SPCM. In the electrical characteristics, a corresponding tran-sition is observed from the linear resistance of a transistor channel to the non-linear rectification of an ideal diode. The high degree of control of ideal p-n junctions using the local gates, combined with the precise photocurrent imaging of the p-n junction position, demonstrate the potential of carbon

nanotubes and the SPCM technique in optoelectronics applications.

ACKNOWLEDGMENTS

This research was supported by a Marie Curie Intra European Fellowship within the 7th European Community Framework Programme, a FOM projectruimte, and NWO Veni and Vidi programs.

1_{J.-C. Charlier, X. Blase, and S. Roche,Rev. Mod. Phys.}

79, 677 (2007).

2

J. U. Lee, P. P. Gipp, and C. M. Heller,Appl. Phys. Lett.85, 145 (2004).

3

J. U. Lee,Appl. Phys. Lett.87, 073101 (2005).

4_{K. Bosnick, N. M. Gabor, and P. L. McEuen,Appl. Phys. Lett.89, 163121}

(2006).

5

N. M. Gabor, Z. Zhong, K. Bosnick, J. Park, and P. L. McEuen,Science 325, 1367 (2009).

6_{T. Mueller, T. Kinoshita, M. Steiner, V. Perebeinos, A. A. Bol, D. B.}

Farmer, and P. Avouris,Nature Nanotech.5, 27 (2010).

7

C.-H. Liu, C.-C. Wu, and Z. Zhong,Nano Lett.11, 1782 (2011).

8

G. A. Steele, G. Gotz, and L. P. Kouwenhoven,Nature Nanotech.4, 363 (2009).

9_{S. M. Sze,Physics of Semiconductor Devices (John Wiley & Sons, New}

York, 1981).

10

A. Imamoglu and Y. Yamamoto,Phys. Rev. B46, 15982 (1992).

11

A. Ho¨gele, C. Galland, M. Winger, and A. Imamoglu, Phys. Rev. Lett. 100, 217401 (2008).

12

J. Kong, H. T. Soh, A. M. Cassell, C. F. Quate, and H. Dai,Nature395, 878 (1998).

13_{Y. H. Ahn, A. W. Tsen, B. Kim, Y. W. Park, and J. Park,Nano Lett.7,}

3320 (2007).

14

K. Balasubramanian, M. Burghard, K. Kern, M. Scolari, and A. Mews, Nano Lett.5, 507 (2005).

15_{S. Rosenblatt, Y. Yaish, J. Park, J. Gore, V. Sazonova, and P. L. McEuen,} Nano Lett.2, 869 (2002).

16

T. W. Odom, J. L. Huang, P. Kim, and C. M. Lieber,Nature391, 62 (1998).

17

For this device, a quantitative estimation of the bandgap out of the transfer characteristic is only possible in vacuum. Indeed, under atmospheric con-ditions, a water-oxygen layer present on the SiO2will screen the gates at

positive voltage and prevent an electron current to flow in the nanotube C. M. Aguirre, P. L. Levesque, M. Paillet, F. Lapointe, B. C. St-Antoine, P. Desjardins, and R. Martel,Adv. Mater.21, 3087 (2009).

18

The difference in the resistance here is due to a combination of hysteresis and inaccuracy in the linear fit due to too few measurement points in this narrow VSD range.

074308-4 Buchs et al. J. Appl. Phys. 110, 074308 (2011)